lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Thu, 10 Jun 2021 15:14:16 -0700
From:   Dmitry Torokhov <dmitry.torokhov@...il.com>
To:     Matti Vaittinen <matti.vaittinen@...rohmeurope.com>
Cc:     Chanwoo Choi <cw00.choi@...sung.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>,
        Bartlomiej Zolnierkiewicz <b.zolnierkie@...sung.com>,
        MyungJoo Ham <myungjoo.ham@...sung.com>,
        Hans de Goede <hdegoede@...hat.com>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH RESEND v2 3/5] extcon: extcon-max77693.c: Fix potential
 work-queue cancellation race

On Thu, Jun 10, 2021 at 12:57:40PM +0300, Matti Vaittinen wrote:
> 
> On Thu, 2021-06-10 at 18:43 +0900, Chanwoo Choi wrote:
> > On 6/8/21 7:10 PM, Matti Vaittinen wrote:
> > > The extcon IRQ schedules a work item. IRQ is requested using devm
> > > while
> > > WQ is cancelld at remove(). This mixing of devm and manual
> > > unwinding has
> > > potential case where the WQ has been emptied (.remove() was ran)
> > > but
> > > devm unwinding of IRQ was not yet done. It may be possible the IRQ
> > > is
> > > triggered at this point scheduling new work item to the already
> > > flushed
> > > queue.
> > > 
> > > According to the input documentation the input device allocated by
> > > devm_input_allocate_device() does not need to be explicitly
> > > unregistered.
> > > Use the new devm_work_autocancel() and remove the remove() to
> > > simplify the
> > > code.
> > > 
> > > Signed-off-by: Matti Vaittinen <matti.vaittinen@...rohmeurope.com>
> > > Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com
> > > >
> > > Reviewed-by: Hans de Goede <hdegoede@...hat.com>
> > > ---
> > > 
> > > Please note that the change is compile-tested only. All proper
> > > testing is
> > > highly appreciated.
> > > ---
> > >  drivers/extcon/extcon-max77693.c | 17 +++++------------
> > >  1 file changed, 5 insertions(+), 12 deletions(-)
> > > 
> > > diff --git a/drivers/extcon/extcon-max77693.c
> > > b/drivers/extcon/extcon-max77693.c
> > > index 92af97e00828..1f1d9ab0c5c7 100644
> > > --- a/drivers/extcon/extcon-max77693.c
> > > +++ b/drivers/extcon/extcon-max77693.c
> > > @@ -5,6 +5,7 @@
> > >  // Copyright (C) 2012 Samsung Electrnoics
> > >  // Chanwoo Choi <cw00.choi@...sung.com>
> > >  
> > > +#include <linux/devm-helpers.h>
> > >  #include <linux/kernel.h>
> > >  #include <linux/module.h>
> > >  #include <linux/i2c.h>
> > > @@ -1127,7 +1128,10 @@ static int max77693_muic_probe(struct
> > > platform_device *pdev)
> > >  	platform_set_drvdata(pdev, info);
> > >  	mutex_init(&info->mutex);
> > >  
> > > -	INIT_WORK(&info->irq_work, max77693_muic_irq_work);
> > > +	ret = devm_work_autocancel(&pdev->dev, &info->irq_work,
> > > +				   max77693_muic_irq_work);
> > > +	if (ret)
> > > +		return ret;
> > >  
> > >  	/* Support irq domain for MAX77693 MUIC device */
> > >  	for (i = 0; i < ARRAY_SIZE(muic_irqs); i++) {
> > > @@ -1254,22 +1258,11 @@ static int max77693_muic_probe(struct
> > > platform_device *pdev)
> > >  	return ret;
> > >  }
> > >  
> > > -static int max77693_muic_remove(struct platform_device *pdev)
> > > -{
> > > -	struct max77693_muic_info *info = platform_get_drvdata(pdev);
> > > -
> > > -	cancel_work_sync(&info->irq_work);
> > > -	input_unregister_device(info->dock);
> > 
> > I think that you have to keep the input_unregister_device().
> 
> Are you sure? I can add back the remove() if required - but the
> kerneldoc for devm_input_allocate_device() seems to be suggesting that
> this would not be needed:
> 
>  * Managed input devices do not need to be explicitly unregistered or
>  * freed as it will be done automatically when owner device unbinds
> from
>  * its driver (or binding fails). Once managed input device is
> allocated,
>  * it is ready to be set up and registered in the same fashion as
> regular
>  * input device. There are no special devm_input_device_[un]register()
>  * variants, regular ones work with both managed and unmanaged devices,
>  * should you need them. In most cases however, managed input device
> need
>  * not be explicitly unregistered or freed.
> 
> https://elixir.bootlin.com/linux/v5.13-rc5/source/drivers/input/input.c#L1955
> 
> I am not going to argue with you though - I am not really familiar with
> the input subsystem. I'd appreciate if someone could shed some light on
> when the input_unregister_device() can be omitted? 

That is correct, you do not need to call input_unregister_device() for
input devices allocated with devm_input_allocate_device().

Thanks.

-- 
Dmitry

Powered by blists - more mailing lists